Sunk Cost sunkcost.ai Data checked 2026-09-03

Devstral 2 123B vs DeepSeek-R1-Distill-Qwen-32B

Devstral 2 123B scores higher on the intelligence index — 9 against 8. DeepSeek-R1-Distill-Qwen-32B is the smaller download at 20 GB, so it runs on cheaper hardware.

Devstral 2 123BDeepSeek-R1-Distill-Qwen-32B
Intelligence index98
ClassBelow every hosted tierBelow every hosted tier
Weights75 GB20 GB
QuantisationQ4_K_MQ4_K_M
Parameters125B32.8B
Max context256k128k
API price per 1M$0.4 in / $2 out$0.8 in / $0.8 out
Licenceother (see repo)MIT
Cheapest machine that runs itStrix Halo Framework Desktop, 128GB $3,449Strix Halo Framework Desktop, 64GB $1,959
Summarising not rated usable
Translation not rated usable
Everyday coding not rated usable
Reasoning & maths not rated good
Agentic work not rated don’t

Ratings are coarse on purpose. Speeds and pay-back depend on the machine — open either model's page for the full list, or see both against the frontier. Context is 32k throughout.